INTRODUCTION Sheep reproduction is specific compared to the reproduction of other animal species, which creates unique opportunities and challenges in the production of sheep meat, milk and wool. Different breeds of sheep can manifest a wide range of reproductive performances depending on their genetic background and ecological/production system. High reproductive efficiency of prolific breeds of sheep can be achieved with a large number of lambs per litter and/or short interlambing periods. Reproductive efficiency is most often expressed by multiple indicators that can be registered at different stages of life and production. Within breed variations of reproductive performance indicators are caused by genetic and non-genetic factors. Understanding the sources of variation in reproductive performance indicators contributes to the effective reproduction and population improvement. Given the complex interactions of different factors that to a greater or lesser extent can contribute to the variability of a particular reproductive performance indicator, researchers are looking for optimal models that can explain the variability of a indicator in the best possible way. This research is based on the assumptions that genetic and non-genetic factors contribute to the variability of reproductive performance indicators in prolific breeds of sheep and that early reproductive performance indicators can be used as predictors of lifetime lamb production. MATERIAL AND METHODS The study included 134 ewes of the Romanov breed. The data were derived from the total of 742 lambings. All sheep originated from one farm in mountain part of Croatia. Ewes were kept at the semi-intensive production system. Reproductive management was based on the accelerated lambing with continuous mating. Ewe lambs were first joined with rams at the age of four to eight months. After joining, rams and ewe lambs were kept together permanently. The breeding group consisted of one ram and approximately 35 to 40 ewe lambs. Reproductive performance indicators included in the analysis were divided into early indicators (age at 1st lambing, 1st litter size and 1st interlambing period) and lifetime production expressed as the total number of lambs born within the first four years of ewes’ life. Linear models for the analysis of variability of reproductive performance indicators included following non-genetic and genetic factors: month of birth, mating group, type of first mating, ram, month of first lambing, type of the first litter, mother – daughter relationship, growth differentiation factor 9 gene (GDF9) and insulin-like growth factor 1 (IGF1) gene. Polymorphisms of GDF9 and IGF1 genes were determined by polymerase chain reaction (PCR) followed by restriction with Hha1 and HaeII restriction enzymes. For the analysis of lifetime production, ewes were divided into several groups in regard to the early reproductive performance indicators and birth weight of lambs at 1st lambing. Thus, the model included age at 1st lambing, 1st litter size, 1st interlambing period, GDF9 and IGF1 genotypes, birth weight of lambs at 1st lambing and mother – daughter relationship. All variables included in the model were categorical. The lowest value of Akaike information criterion (AIC) was used to determine the optimal model for the prediction of the lifetime production. Phenotypic and genotypic correlations were calculated, depending on the type of the variable, with Pearson and Spearman rank correlations. Heritability was estimated by mother – daughter regression. RESULTS By the age of four, the examined sheep lambed between 4 and 6 times. The mean age at first lambing was 373.2 ± 34.5 days, with the range from 304.0 to 469.0 days. Non-genetic factors with the greatest influence on this trait were type of first mating (F(2,123)=130.25; P>0.001) and month of birth (F(4,123)=45.72; P>0.001). Significant differences (P<0.05) in age at 1st lambing were observed between ewes with different types of the first mating and between ewes born in different months. Genotypic correlations of age at 1st lambing with size of the 1st litter and 1st interlambing period were -0.12 and 0.01, respectively. Phenotypic correlations between these traits were very weak. Estimated heritability of age at 1st lambing was 0.38 (P=0.360). The average size of the 1st litter was 1.98 ± 0.54 lambs. The largest number of ewes had twins at first lambing (95), while 21 ewe had singleton and 18 had triplets. No significant differences (P>0.05) were observed between frequencies of singletons, twins and triplets in ewes divided in groups according to the ram, type of the first mating, month of birth, mother – daughter relationship, GDF9 genotypes and IGF1 genotypes. Model which included previously listed non-genetic and genetic factors explained relatively small proportion of the variance of this indicator (R2=0.0764; F(14,119)=0.703; P=0.768). Phenotypic and genotypic correlation of 1st litter size and 1st interlambing period were 0.19 and 0.04, respectively. Heritability of the 1st litter size was 0.20 (P=0.694). The 1st interlambing period lasted an average of 257.8 ± days, with the range from 196.0 to 393.0 days. The largest contribution to the variations in this indicator had the factor of the breeding group (F(3,121)=19.26; P<0.001). Ewes in the fourth breeding group had significantly (P<0.05) longer 1st interlambing period (309.9 ± 8.5) compared to the ewes in breeding group one (261.2 ± 6.8), two (248.7 ± 7.5) and three (242.6 ± 7.5). Significant differences (P<0.05) in 1st interlambing period were also observed between ewes born in January (257.3 ± 3.4) and April (269.2 ± 8.7) and between ewes whose first lambing occurred in February (270.1 ± 8.8) and March (255.9 ± 5.6). Estimated heritability was -0.42 (P>0.398). The average lifetime production of the investigated sheep was 11.87 ± 2.33 lambs, with the range from 7 to 18 lambs. The model explained 25.49 % of the variations in lifetime production (R2=0.2549; F(16,117)=2.43; P<0.001). The greatest contribution to the individual variations in this indicator had size of the 1st litter (F(2,117)=2.77; P=0.047) and 1st interlambing period (F(2,117)=5.12; P=0.007). Significant differences (P<0.05) in lifetime production were observed between ewes that lambed triplets in first litter (13.43 ± 0.94) compared to ewes that lambed twins (11.90 ± 0.82) and singletons (9.60 ± 1.12). Ewes with short 1st interlambing period had significantly (P=0.001) larger lifetime production (12.52 ± 0.58) compared to those with long 1st interlambing period (10.79 ± 0.61). Genotypic correlations between age at 1st lambing, 1st litter size, 1st interlambing period and lifetime production were -0.01, 0.18 and -0.22 respectively. Phenotypic correlations between these traits were -0.13, 0.34 and -0.28, respectively. Heritability of lifetime production was 0.21 (P=0.521). The optimal model for the prediction of lifetime production included size of the 1st litter and the 1st interlambing period as independent variables, with an estimated precision of 15.86 %. CONCLUSIONS The studied population of Romanov ewes showed the ability for frequent lambing. The average litter size values were at the level expected for the prolific breeds. Intrabreed variations in reproductive performance indicators related to the number of lambs born (size of the 1st litter and lifetime production) were higher compared to the age at 1st lambing and 1st interlambing period. The age at 1st lambing varied the most depending on the month of birth and the type of mating. Non-genetic factors related to the age of sheep and the type of mating, as well as genetic factors of GDF9 and IGF1 gene polymorphisms and kinship, did not significantly explain the variations in the size of the 1st litter. Variations in the 1st interlambing period are mostly explained by the factor of the breeding group, which included the influence of ram and social interactions between individuals. The size of the 1st litter and the 1st interlambing period were factors with the greatest contribution to variation in lifetime produciton. Ewes with higher number of lambs in first litter and with a shorter 1st interlambing period had higher lifetime production. The optimal model for predicting lifetime production included the size of the 1st litter and the 1st interlambing period as independent variables. Phenotypic and genotypic correlations between age at 1st lambing, size of the 1st litter, 1st interlambing period and lifetime production of these indicators have little applicability in breeding and selection schemes aimed to improve sheep fertility.